European Journal of Echocardiography 2006 7(5):356-364; doi:10.1016/j.euje.2005.08.004
Copyright © 2005, The European Society of Cardiology
Disagreement between tissue Doppler imaging and conventional pulsed wave Doppler in the measurement of myocardial performance index
Eduardo Casas Rojo*,
José Luis Rodrigo,
Leopoldo Pérez de Isla,
Carlos Almería,
Nieves Gonzalo,
Adalía Aubele,
Rafael Cinza,
José Zamorano and
Carlos Macaya
Hospital Clínico San Carlos, Instituto Cardiovascular, C/Professor Martín Lagos s/n, 28040 Madrid, Spain
Received 27 April 2005; received in revised form 8 August 2005; accepted after revision 19 August 2005.
* Corresponding author. Tel.: +34 91 669286904; fax: +34 91 3303293. ecasasweb{at}jazzfree.com
 |
Abstract
|
|---|
Aim Myocardial performance index (MPI) is usually measured with
pulsed wave Doppler (PWD). Our aim was to assess the degree
of agreement between PWD and a method based on tissue Doppler
imaging (TDI).
Methods and results Seventy-five patients with prior myocardial infarction and 20 healthy subjects underwent measurement of time intervals and MPI with PWD and pulsed TDI at septal and lateral sides of mitral annulus. MPI and TDI-MPI at septal side showed the best intraclass correlation coefficient (ICC=0.54; p<0.0005). Ninety-five percent interval of agreement ranged from –0.27 to 0.22. These differences were attributed to discrepancies in isovolumic contraction and relaxation times. In the healthy group the results were similar (ICC=0.44), although the 95% interval of agreement was lower (from –0.13 to 0.12).
Conclusions The agreement between PWD and TDI in the measurement of MPI is only moderate. This should be taken into account in the interpretation of studies in which TDI is used for this measurement.
Keywords: Tissue Doppler imaging; Echocardiography; Tei index; Myocardial performance index
|
Introduction
|
|---|
Myocardial performance index (MPI), defined as the sum of isovolumic
contraction and relaxation times divided by ejection time, is
a powerful index which provides evaluation of systolic and diastolic
functions at the same time.
1,2 Its prognostic value in patients
with diverse heart diseases such as myocardial infarction has
been proved in several studies.
3–15 The classical approach
to measure MPI is pulsed wave Doppler.
2 Recently some investigators
have measured MPI with tissue Doppler imaging (TDI) in substitution
of the classical approach. Our aim was to establish the degree
of agreement between classical and TDI methods in patients with
a recent myocardial infarction.
|
Methods
|
|---|
Seventy-seven patients featuring a prior myocardial infarction
(MI) less than a year before were included in this study and
underwent a complete Doppler-echocardiographic study. Time intervals
were measured in 75 of them (97.4%), as 2 patients were excluded
because they did not have a sinusal cardiac rhythm (one of them
showed atrial fibrillation, the other showed pacemaker rhythm).
Thirty-seven of these patients underwent a second echocardiographic
study (12.2±18.7 days after the previous one) to assess
test–retest variability of the indexes. A second group
of 20 healthy young people also underwent a similar study in
order to evaluate the relationship between classical Doppler
and TDI intervals in the absence of myocardial disease.
|
Echocardiography
|
|---|
Fig. 1 shows the Doppler images recorded for the measurement
of time intervals necessary to calculate MPI with conventional
and TDI methods.
View larger version (86K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Figure 1 Time intervals measured from pulsed Doppler transmitral inflow and left ventricular outflow (PEP=preejection period, which includes EMD=electromechanical delay and ICT=isovolumic contraction time; ET=ejection time; IRT=isovolumic relaxation time) and their equivalent intervals measured from pulsed tissue Doppler (tPEP=tissue preejection period; tEMD=tissue electromechanical delay; tICT=tissue isovolumic contraction time; tET=tissue ejection time; tIRT=tissue isovolumic contraction time). MPI is calculated as (a–b)/b which is equal to (ICT+IRT)/ET; tMPI is calculated as (tICT+tIRT)/tET.
|
|
Conventional Doppler
All the studies were performed with a Sonos 4500 equipment (Andover,
USA) with a 2.5-mHz transducer. Pulsed Doppler study of transmitral
inflow was made placing the sample between the tips of the mitral
leaflets in the four chambers' view and "
a" interval was measured
between cessation and onset of the mitral inflow. Pulsed Doppler
study of left ventricular (LV) outflow was made placing the
sample just below the aortic valve in the five chambers and
"
b" interval was measured between onset and cessation of the
LV outflow. Intervals "
a" and "
b" were obtained from the average
of three consecutive cardiac cycles. Conventional MPI was obtained
as (
a–
b)/
b.
Other intervals were also measured from the same recordings: preejection period (PEP) was defined as the interval from the onset of Q wave of ECG to the onset of LV outflow; electromechanical delay (EMD) from Q wave to cessation of mitral inflow; isovolumic contraction time (ICT) was calculated as (PEP–EMD); isovolumic relaxation time (IRT) was obtained as (a–b–ICT). Finally, filling time (FT) was calculated as RR–a, where RR is the interval in milliseconds, deduced from heart rate. All measurements were performed off-line on digital recordings.
TDI
Pulsed wave TDI was performed at septal mitral annulus and lateral mitral annulus. Three consecutive cycles were recorded in order to obtain the mean values. TDI isovolumic contraction time (tICT) was measured between cessation of A' wave and onset of S wave; TDI ejection time (tET) was obtained between onset and cessation of S wave; TDI isovolumic relaxation time (tIRT) was obtained between cessation of S wave and onset of E' wave. tMPI was calculated as (tICT+tIRT)/(tET). Three types of tMPI were defined: tMPI at septal annulus (tMPIs), at lateral annulus (tMPIl) and the mean value of both (tMPIm).
|
Statistics
|
|---|
Intraclass correlation coefficient (with absolute agreement
definition) (ICC) was used to assess the degree of agreement
between conventional and TDI methods. ICC<0.40 was considered
mild agreement; 0.41–0.75 moderate; 0.76–1 good
agreement.
14,15 Bland and Altman plots were used to show the
intervals of agreement between the methods, in order to evaluate
the clinical relevance of the differences.
16 Pearson's correlation
coefficient was used to assess linear correlation between the
new indexes and other clinical and echocardiographic variables.
A paired
t-test was used to evaluate the differences between
several time intervals measured by the different methods. A
p value below 0.05 was considered statistically significant.
Variability was defined as the difference between the measurements divided by the mean value of both and is expressed as percentage. Interobserver and intraobserver variabilities were calculated with 2 measurements from the same recordings (measured by two observers for interobserver variability, and measured by the same observer for intraobserver variability). Test–retest variability (reproducibility) was calculated with measurements from two different studies of the same patients (measured by the same observer).
|
Results
|
|---|
Patients with prior MI
The characteristics of this group of patients are displayed
in
Table 1. Doppler measurements with conventional and new TDI
methods are shown in
Table 2. The global differences between
MPI and tMPI were not significant (
p=0.057 for MPI and tMPIs;
p=0.16 for MPI and tMPIl). Of the two regions of interest explored
with TDI, the most reliable index was tMPI measured at septal
mitral annulus (tMPIs). ICC for tMPIs (95% confidence interval
ICC=0.36–0.69) suggested a significant but only mild to
moderate degree of agreement between tMPI and conventional MPI.
ICC was lower for tMPIl (95% CI ICC=0.09–0.49) and tMPIm
(0.28–0.63).
Fig. 2 shows Bland and Altman plot of the
differences between both the methods in relation to the average
value. There was a clinically important disagreement in many
of the MPI estimations (95% limits of agreement from –0.22
to 0.27). Twenty patients (26.7%) showed a difference between
methods higher than 0.1. Further details about Bland and Altman
analysis of these patients are shown in
Fig. 3 and
Table 3.
When analyzing only patients without defects of contractility
in basal segments (
n=29), the results were similar (Bland and
Altman 95% interval of agreement from –0.27 to 0.22; ICC=0.35;
p=0.03).
View larger version (13K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Figure 2 Bland and Altman plots of MPI and tMPI at septal annulus (tMPIs), lateral annulus (tMPIl) and the mean values of both (tMPIm) for the MI group. The differences between the methods are displayed against the average of both measures. The lines represent mean difference and 95% interval of agreement in absolute value of the differences.
|
|
View larger version (25K):
[in this window]
[in a new window]
[Download PowerPoint slide]
|
Figure 3 Bland and Altman plots for the components of MPI in the MI group: isovolumic contraction time (ICT), ejection time (ET) and isovolumic relaxation time (IRT). On the left, comparisons between the conventional Doppler method and TDI method at septal annulus. On the right, the same comparisons for lateral annulus.
|
|
The analysis of the mean differences of time intervals showed
that there were significant differences in the measurement of
all of them (PEP, ICT, ET, IRT, FT) between the two methods
(conventional and TDI), with the exception of EMD (NS). Of the
intervals needed to define MPI, ejection time (ET) estimations
showed the best intraclass correlation (ICC=0.7188;
p<0.0005).
ICT and IRT only showed mild agreement between the methods.
There were no significant linear correlations between any of MPI measurement methods and heart rate or systolic/diastolic arterial pressure values. Mild correlations of MPI and tMPI with left ventricular ejection fraction (LVEF) and wall motion score index (WMSI) are displayed in Table 4.
Inter and intraobserver variabilities
Interobserver variability was 7.6%±4.3% for MPI; 18.8%±21.5%
for tMPIs; 21.5%±21.4% for tMPIl. Intraobserver variability
was 8.3%±4% for MPI; 7.8%±11% for tMPIs; 9.3%±10.1%
for tMPIl.
Test–retest variability
In the subgroup undergoing a second study for test–retest comparisons (n=37) there were no significant differences between both the studies in heart rate (66.38±11.18 vs. 67.43±12.57) or diastolic blood pressure (83.06±13.02 vs. 82.26±14.31). There was a statistically non-significant trend to higher systolic blood pressure values in the second study (130±16.43 vs. 138.71±28.84; p=0.06). Mean MPI values were similar in both studies (0.46±0.11 vs. 0.45±0.11) but there were significant differences in tMPIs (0.43±0.14 vs. 0.47±0.13; p=0.04) and tMPIl (0.41±0.13 vs. 0.46±0.13).
The values of test–retest variability in these patients were 16.9%±16.9% for MPI; 22.1%±19.4% for tMPIs; 18.6%±20% for tMPIl.
Healthy control group
Time intervals and indexes measured with conventional method and TDI methods in this group are shown in Table 5.The values of MPI ranged from 0.27 to 0.46. Both tMPIs and tMPIl showed a mild to moderate intraclass correlation with MPI (95% CI=0.00–0.73 for ICC between MPI and tMPIs; 0.09–0.76 for ICC between MPI and tMPIl; 0.24–0.82 for ICC between MPI and tMPIm). Fig. 4 shows Bland and Altman plots showing the 95% agreement interval. More details are displayed in Table 6.
|
Discussion
|
|---|
Although most studies refer to TDI's focus on the measurement
of velocities, some investigators have used TDI to obtain systolic
and diastolic time intervals.
17 These can be measured from pulsed
wave TDI or color M-Mode TDI recordings. Pulsed wave TDI has
the advantage of a high temporal resolution, and it only requires
good visualization of the region of interest selected, usually
mitral annulus. Conventional measurement of MPI usually requires
the recording of transmitral flow and left ventricular outflow
separately and therefore its calculation is relatively complex
because of the need to obtain an average measure of several
cardiac cycles, in order to avoid the influence of heart rate
variations. The measurement of time intervals from pulsed wave
TDI could provide an easier way to calculate MPI and other parameters
within a single cardiac cycle. There has been few attempts to
validate TDI as a method for measuring MPI.
18–22 According
to these studies the agreement between the methods is high in
healthy subjects and dilated cardiomyopathy, and slightly lower
in patients with previous myocardial infarction (MI). The main
limitation of these studies is the very low number of patients
included. Our study shows that MPI measured with mitral annulus
tissue velocity intervals has only a mild to moderate agreement
with traditional MPI.
In patients with prior myocardial infarction, TDI intervals may be influenced by intraventricular conduction disturbs, asynchrony, and the differences in the contraction and relaxation times between the different myocardial segments. Mitral annulus velocity intervals, in particular, could be highly influenced by the contractility of basal segments. However, the differences between conventional MPI and tMPI were similar in patients with good contractility of basal segments and even in patients without segmentary contractility alterations. Furthermore, similar observations can be found in healthy subjects. Therefore the disagreement between methods does not seem to depend on contractility alterations. We observed that the values of ICT and ET intervals measured with TDI were larger than the ones obtained with conventional Doppler; this could increase tMPI values. On the other hand, IRT interval was usually shorter with TDI, and this could decrease tMPI values. The results of both the influences are values of MPI and tMPI of similar range, but not accurate ones. Of the intervals needed to calculate tMPI, ET showed lower differences with the conventional method, while tICT and tIRT correlated poorly with classic intervals. Therefore, isovolumic times are in most part the responsible components of MPI which explain the differences between conventional MPI and tMPI.
Considering these results, tMPI should be defined as a different index, and not as an alternative way to obtain classic MPI, because tMPI is obtained from tissue velocity time intervals that are not exactly coincident with flow time intervals. This does not mean that tMPI is not useful, but the different cutoff values must be defined, and specific studies must be done to know its features. The best agreement is reached when measuring at septal annulus. This can be partially explained by the worst image definition provided by lateral annulus, and also by the asynchrony between both the locations of mitral annulus. However, in healthy people lateral annulus seemed to be equal or slightly better than septal annulus. The use of mean values of septal and lateral tMPI did not improve the agreement. The analysis of time intervals suggests that systolic velocity time intervals measured with TDI last more time than their correspondent systolic flow intervals (measured with conventional Doppler). These include PEP (divided in subintervals EMD and ICT) and ET. tPEP and its subintervals last more time than conventional ones, and so ejection velocities are delayed. tET is also slightly longer than conventional one, and therefore isovolumic relaxation velocities are very delayed when compared with conventional IRT. On the other side, diastolic velocity intervals seem to be shorter than diastolic flow intervals, especially IRT, and also FT in a minor way. Of these observations we can deduce that left ventricular contraction does not end before aortic valve is closed, moreover it seems to persists a little time after that and therefore relaxation velocities are delayed and shortened when compared with IRT and filling time measured by conventional Doppler. This delay is increased in the presence of myocardial ischemia, as it has been proved by García-Fernández et al.23 and Rivas-Gotz et al.24 but it does exist in patients without regional contractility alterations, and even in healthy people.
With regard to variability data, we found that variability of measuring conventional MPI from the same recordings was little and it did not depend on the observer. On the other hand, intraobserver variability of tMPI was also little but interobserver variability was much bigger. This may be explained because the limits of the different intervals from DTI are often poorly defined and they require a specifically trained observer. Therefore conventional MPI has also the advantage of being easily measured by an average echocardiographist.
Test–retest data show that MPI and tMPI are not accurately reproducible when they are measured two times on different days, specially in the case of tMPI which actually showed significantly higher values in the second study. This may be due to different load conditions and to the higher systolic blood pressure values observed at the second time. Therefore neither MPI nor tMPI seem to be accurate indexes to perform serial studies of the same patients, and particularly tMPI seems to be too sensitive to the mild hemodynamic shifts usually observed in this context.
Knowing that MPI and tMPI showed only a moderate degree of agreement in ischemic patients, data from healthy subjects confirmed that disagreement also in healthy people. In Bland and Altman analysis the differences observed were lower when compared with the study in patients with prior myocardial infarction. This is partially explained by the shorter values and lower range of values observed in healthy people, resulting in lower differences. However, Tables 3 and 6
display the agreement intervals as percentage (which is less influenced by the range of values) showing that the differences observed in healthy people are proportionally lower than the ones observed in ischemic patients. Therefore, the disagreement between methods does not depend entirely on the alterations induced by ischemic disease on left ventricle, but the range of the differences is clinically more important in patients with ischemic disease. In healthy individuals TDI at lateral annulus showed a higher mean ICC with conventional method than TDI at septal annulus did, but this does not seem to be relevant, as the 95% confidence intervals of ICC were wide and similar for both locations.
|
Conclusion
|
|---|
MPI can be easily measured with TDI at mitral annulus, specially
at septal side. However, the agreement between TDI and conventional
method is not good, because systolic intervals are longer and
diastolic intervals are shorter when measured with TDI. The
disagreement exists on healthy people and it is increased on
patients with prior myocardial infarction. This should be taken
into account when interpretating studies in which TDI is used
to measure this index. Further studies are needed to assess
if tMPI has any independent value in the evaluation and prognosis
of heart failure.
|
References
|
|---|
- Mancini G.B., Costello D., Bhargava V., Lew W., LeWinter M., Karliner J.S. The isovolumic index: a new noninvasive approach to the assessment of left ventricular function in man. Am J Cardiol (1982) 50:1401–1408.[CrossRef][ISI][Medline]
- Tei C. New non-invasive index for combined systolic and diastolic ventricular function. J Cardiol (1995) 26:135–136.[ISI][Medline]
- Tei C., Ling L., Hodge D.O., Bailey K.R., Oh J.K., Rodeheffer R.J., et al. New index of combined systolic and diastolic myocardial performance: a simple and reproducible measure of cardiac function – a study in normals and dilated cardiomyopathy. J Cardiol (1995) 26:357–366.[Medline]
- Tei C., Nishimura R.A., Seward J.B., Tajik A.J. Noninvasive Doppler-derived myocardial performance index: correlation with simultaneous measurements of cardiac catheterization measurements. J Am Soc Echocardiogr (1997) 10:169–178.[CrossRef][ISI][Medline]
- Tei C., Dujardin K.S., Hodge D., Kyle R.A., Tajik A.J., Seward J.B., et al. Doppler index combining systolic and diastolic myocardial performance: clinical value in cardiac amyloidosis. J Am Coll Cardiol (1996) 28(3):658–664.[Abstract]
- Dujardin K.S., Tei C., Yeo T.C., Hodge D.O., Ross A., Seward J.B., et al. Prognostic value of a Doppler index combining systolic and diastolic performance in idiopathic-dilated cardiomyopathy. Am J Cardiol (1998) 82:1071–1076.[CrossRef][ISI][Medline]
- Poulsen S.H., Jensen S.E., Nielsen J.C., Moller J.E., Egstrup K. Serial changes and prognostic implications of a Doppler-derived index of combined left ventricular systolic and diastolic performance in acute myocardial performance in acute myocardial infarction. Am J Cardiol (2000) 85:19–25.[CrossRef][ISI][Medline]
- Poulsen S.H., Jensen S.E., Tei C., Seward J.B., Egstrup K. Value of the Doppler index of myocardial performance in the early phase of acute myocardial infarction. J Am Soc Echocardiogr (2000) 13:723–730.[CrossRef][ISI][Medline]
- Moller J.E., Soendergaard E., Poulsen S.H., Appleton C.P., Egstrup K. Serial Doppler echocardiographic assessment of left ventricular performance after a first myocardial infarction. J Am Soc Echocardiogr (2001) 14:249–255.[CrossRef][ISI][Medline]
- Peltier M., Slama M., Garbi S., Enriquez-Sarano M.L., Goissen T., Tribouilloy C.H. Prognostic value of Doppler-derived myocardial performance index in patients with left ventricular systolic dysfunction. Am J Cardiol (2002) 90:1261–1263.[CrossRef][ISI][Medline]
- Moller J.E., Egstrup K., Kober L., Poulsen S.H., Nyvad O., Torp-Pedersen C. Prognostic importance of systolic and diastolic function after acute myocardial infarction. Am Heart J (2003) 145:147–153.[CrossRef][ISI][Medline]
- Lavine S.J. Prediction of heart failure post-myocardial infarction: comparison of ejection fraction, transmitral filling parameters, and the index of myocardial performance. Echocardiography (2003) 20(8):691–701.[CrossRef][ISI][Medline]
- Ascione L., De Michele M., Accadia M., Rumolo S., Damiana L., Guarini P., et al. Myocardial global performance index as a predictor of in-hospital cardiac events in patients with first myocardial infarction. J Am Soc Echocardiogr (2003) 16:1019–1023.[CrossRef][ISI][Medline]
- Mc Graw K.O., Wong S.P. Forming inferences about some intraclass correlation coefficients. Psychol Methods (1996) 1:30–46. [Correction: 1:390].[CrossRef][ISI]
- Fleiss J.L. The design and analysis of clinical experiments. (1986) New york: Wiley.
- Bland M., Altman D.G. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet (1986) 1:307–310.[CrossRef][ISI][Medline]
- Zamorano J., Wallbridge D.R., Ge J., Drozd J., Nesser J., Erbel R. Non-invasive assessment of cardiac physiology by tissue Doppler echocardiography. Eur Heart J (1997) 18:330–339.[Abstract/Free Full Text]
- Harada K., Tamura M., Toyono M., Yasuoka K. Comparison of the right ventricular Tei index by tissue Doppler imaging to that obtained by pulsed Doppler in children without heart disease. Am J Cardiol (2002) 90:1173–1178.
- Pellerin D., Larrazet F., Cohen L., Witchitz S., Veyrat C. Myocardial time intervals preceding left ventricular filling in chronic coronary artery disease: value of a decreased septal ejection time. Int J Cardiol (2003 May) 89(1):33–44.[CrossRef][ISI][Medline]
- Tekten T., Onbasili A., Ceyhan C., Unal S., Discigil B. Value of measuring myocardial performance index by tissue Doppler echocardiography in normal and diseased heart. Jpn Heart J (2003 May) 44(3):403–416.[CrossRef][Medline]
- Tekten T., Onbasili A., Ceyhan C., Unal S., Discigil B. Novel approach to measure myocardial performance index: pulsed-wave tissue Doppler echocardiography. Echocardiography (2003) 20(6):503–510.[CrossRef][ISI][Medline]
- Baykan M., Yilmaz R., Celik S., Orem C., Kaplan S., Erdol C. Assessment of left ventricular systolic and diastolic function by Doppler tissue imaging in patients with preinfarction angina. J Am Soc Echocardiogr (2003) 16:1024–1030.[CrossRef][ISI][Medline]
- García-Fernández M.A., Azevedo J., Moreno M., Bermejo J., Perez-Castellano N., Puerta P., et al. Regional diastolic function in ischemic heart disease using pulsed wave Doppler tissue imaging. Eur Heart J (1999) 20:496–505.[Abstract/Free Full Text]
- Rivas-Gotz C., Khoury D.S., Manolios M., Rao L., Kopelen H.A., Nagueh S.F. Time interval between onset of mitral inflow and onset of early diastolic velocity by tissue Doppler: a novel index of left ventricular relaxation. J Am Coll Cardiol (2003) 42(8):1463–1470.[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
Top
Abstract
Introduction